Intelligent Sentiment-Driven Recommendation System for E-Commerce using Deep Learning and Hybrid Optimization

Abstract

Personalized recommendation systems have become essential in modern e-commerce platforms due to the rapid growth of online products and user-generated reviews. Traditional recommender systems primarily focus on user–item interactions and often fail to capture the emotional and opinion-based information hidden within customer reviews. To address this limitation, this research proposes a deep sentiment-aware recommendation framework that integrates sentiment analysis with personalized recommendation strategies to enhance recommendation quality and user engagement. The proposed system combines deep learning-based sentiment prediction, collaborative filtering mechanisms, and hybrid optimization techniques to generate accurate and context-aware recommendations. Customer reviews are analyzed using advanced natural language processing and deep neural architectures to extract sentiment polarity and semantic relationships. These sentiment features are then integrated with user interaction history to improve recommendation relevance and ranking performance. Furthermore, an optimization-driven recommendation pipeline is employed to enhance recommendation precision, reduce sparsity issues, and improve scalability for large-scale datasets. The system utilizes Amazon product review datasets containing user ratings, review texts, and interaction logs for experimental evaluation. Experimental outcomes demonstrate that the proposed sentiment-aware recommendation framework significantly improves recommendation effectiveness, personalization quality, and customer satisfaction compared to conventional recommendation models.